Heuristics and Biases? Laplace was there, 200 years ago.

In an article entitled Laplace’s Theories of Cognitive Illusions, Heuristics, and Biases, Josh “hot hand” Miller and I write:

In his book from the early 1800s, Essai Philosophique sur les Probabilités, the mathematician Pierre-Simon de Laplace anticipated many ideas developed in the 1970s in cognitive psychology and behavioral economics, explaining human tendencies to deviate from norms of rationality in the presence of probability and uncertainty. A look at Laplace’s theories and reasoning is striking, both in how modern they seem and in how much progress he made without the benefit of systematic experimentation. We argue that this work points to these theories being more fundamental and less contingent on recent experimental findings than we might have thought.

We conclude:

Laplace’s approach to identifying behavior that departed from the enlightenment conception of rational decision making—an effort that occurred in parallel with his role as a major architect of this ideal, as it applied to inference and decision making under uncertainty—spurred him to search for the general principles of reasoning that underlay these departures. That many of his explanations happen to coincide with modern accounts, arrived at independently based on the same introspections that evidently guided Laplace, suggests that the heuristics and biases approach to judgement and decision making is a scientific contribution that will endure.

More generally, Laplace’s work as a proto-psychologist and applied statistician, which complemented his career as a mathematician and physicist, demonstrates the creative tension between normative and descriptive ideas of inference and decision making. . . .

Modern behavioral science research has taken us far beyond Laplace. While Laplace was an early advocate for the scientific method to be applied to psychological questions, he was limited in his inquiry by his reliance upon observational data. Modern research, through the use of innovative and carefully designed experimental demonstrations, has provided insights and further directions of study into how and why human behavior departs from the normative model of probability theory (Kahneman et al., 1982). Looking at decision making from a different direction, as Laplace’s faith in a clockwork universe that could be reduced to intelligible causes via the scientific method has been called into question with the discovery of quantum phenomena and emergent complexity, Laplace’s assumption that probability theory could serve as a domain-independent prescriptive model for human judgement has been upended by research demonstrating the relative efficacy of simple domain-specific decision rules and predictive models that respect cognitive limitations, tacit knowledge, multidimensionality of goals, and the need to adapt to complex and changing environments (Meehl, 1954; Gigerenzer and Brighton, 2009; Todd and Gigerenzer, 2000).

Nevertheless, Laplace’s attempts to understand the underlying mechanisms for people’s biases were highly original, insightful, in many ways were centuries ahead of their time, and in at least two instances produced novel conjectures that have not been tested to this day. We believe that modern-day social and behavioral scientists can benefit from revisiting Laplace’s thinking on illusions in the estimation of probabilities, and beyond.

P.S. Emma Gillingham sent in the above photo of Pepper, along with the following description:

The classic ‘in or out’ debate – open a door, the cat wants to stay in. Close it, the cat wants to go out. Repeat for the next few hours. Some kind of cognitive bias that the grass is always greener?!

67 thoughts on “Heuristics and Biases? Laplace was there, 200 years ago.

    • It’s obvious to me that Gillingham has her own cognitive bias. To my mind/bias, the cat is looking to make sure that her options aren’t constrained by some silly human.

      • Maybe it’s just a game that the cat likes to play.
        Or maybe the cat’s desire to go out is contingent on something else that it can’t express. (What makes me think of this: The dog my family had as a kid would bark at the door, sounding as if he wanted to go out. So I let him out. Then he sat outside the door, barking as if he wanted to come in. But it wasn’t in or out that he wanted — it was out with me. He was fine outside if I just sat in a chair outside reading.)

        • Ahem…

          Time to recall (or state) Colette’s axiom : “A cat is almost surely on the wrong side of any door”.

          The corollaries of this axiom in quantum mechanics have not yet be considered. What a pity…

  1. I wonder who will respond to your article. Looks super interesting. My favorite topic these days. I do think that the over 256 or more biases so identified have to be revisited for lack of clarity and to standardize their definitions and meaning. I’ve been mulling over this. And see that there is a project in the making in this regard.

  2. Fascinating about Laplace. I think however we have even earlier sources relating human psychological biases to distortion of scientific observations and claims, e.g., in Francis Bacon’s writings about “idols”
    https://en.wikipedia.org/wiki/Novum_Organum#The_Idols_(Idola)
    – that is of course pre-probabilistic but still applies to general scientific reasoning.
    Of course the earlier we go, the more effort is needed for translation into modern terms and categories, but as Miller and Gelman’s paper shows the effort in resurrecting very old observations is worthwhile.

    • Sander,

      Indeed, on the cutting room floor we have this quote from Francis Bacon on the primacy effect, which can only be over-come by what Andrew calls the time reversal heuristic:

      the first conclusion colors and brings into conformity with itself all that come after.

      Also, there is some nice related work by Ashraf, Camerer, and Loewenstein on how Adam Smith anticipated some of the (non-probabilistic) content of Behavioral Economics. Hume has a bunch too.

    • “Come, let us cast lots to find out who is responsible for this calamity.”

      The road from asking that lots decide, to understanding what patterns they may conceal or reveal, has been a long one.

      As for Josh and Andrew’s paper, fun stuff. As a lawyer I’m especially intrigued by this: “In particular, Laplace leverages its framework in order to develop novel (at the time) insights into how to assess the (posterior) credibility of witnesses to improve decision making in the courtroom, and in daily life.” And, of course, props for the shoutout to our patron saint Cicero – “Probability is the very guide of life”.

      • My only puzzlement has been is what is it with examples of ‘drawing white or red balls’ from urns toward illustrating specific biases & decision proclivities.

        I read Howard Raiffa’s and Ralph Keeney’s work back in the late 80’s or early 90’s and wondered whether such examples really advanced decision analytics. If anything I think we rely excessively on fundamental attribution and NHST in so many areas.

        Incidentally Howard Raiffa was the Frank P. Ramsey Professor at Harvard Business School. T

        • Sameera

          When we read the testimony chapter it became clear that Laplace is using Bernoulli’s balls-in-urns framework as a stylized model in order to illustrate precisely what principles eminent legal minds were missing. The neglect of base rates is a big one. I find it useful

          But your point is well taken, John Stuart Mill wrote this very critical piece on Laplace’s application of the probability to the social sciences.

          We have sympathy for it too, in our penultimate paragraph we write:

          Looking at decision making from a different direction, as Laplace’s faith in a clockwork universe that could be reduced to intelligible causes via the scientific method has been called into question with the discovery of quantum phenomena and emergent complexity, Laplace’s assumption that probability theory could serve as a domain-independent prescriptive model for human judgement has been upended by research demonstrating the relative efficacy of simple domain-specific decision rules and predictive models that respect cognitive limitations, tacit knowledge, multidimensionality of goals, and the need to adapt to complex and changing environments (Meehl, 1954; Gigerenzer and Brighton, 2009; Todd and Gigerenzer, 2000).

        • Joshua, Thanks. As I mentioned, I haven’t read Laplace. Just generally, I thought that the framework, balls-in urns, as has been applied in decision analytics had sustained for at least 30 years or more. Thank goodness Tversky & Kahneman went beyond it. I have great respect for Howard Raiffa’s work in negotiation theories.

          Yes the neglect of base rates has also dominated some fields disproportionately. And if the Bernoulli balls-in-urns framework is used as examples to explain base rates, I’m afraid the effort is lost on me. Some people just have intrinsically more fluid and crystallized intelligence. Maybe with some training and exposure to and engagement with exceptional thinking, some of it will sustain provided opportunities are provided to as many people as possible.

      • Thanatos

        Nice. Thanks. The chapter on testimony is a fun read, though I’d recommend doing the calculations yourself rather than following his.

        I didn’t see much more from Cicero on probability, just this from Pascal’s The Science of Conjecture: Evidence and Probability before Pascal:

        Cicero’s account of these matters in his Academica consists of Carneades’ theory minus the best parts. Cicero merely reports Carneades as holding that “the wise man withholds assent” in theory but follows probability in practice, where it is sufficient ground for action. As to what probability is, Cicero is silent, and as to how to discover what is probable, the following is all he offers. While not in itself negligible, it hardly constitutes a sufficient exposition of probability: “When a wise man is going on board a ship, surely he has got the knowledge already grasped in his mind and perceived that he will make the voyage as he intends? How can he have it? But if for instance he were setting out from here to Puteoli, a distance of thirty stadia, with a reliable crew and a good helmsman and in the present calm weather, it would appear probable that he would get there safe.” In expounding his own philosophy in other works, Cicero again approves of advancing doctrines as only probable, but one is left none the wiser as to what this means. In addition to this he inserts into his long works on argument a few remarks of more or less Aristotelian tenor on probability as a property of arguments. Dialectic, he says, “conveys a method that guards us from giving assent to any falsehood or ever being misled by deceitful probability.” Among the things one should consider in constructing arguments is “the nature of the event, whether it is wont to happen commonly or unusually and rarely.

    • Anon:

      Josh was reading that chapter of Laplace’s book (in translation) and noticed the strikingly modern ideas on heuristics and biases. He shared this with me and we wrote the paper together. Then we sent it to some people who offered useful comments and we improved it!

        • That was fortuitous Joshua. Really enjoyed the article. I will probably reread it as is my habit.
          I’ve not read Laplace. I read Tversky and Kahneman’s work when it 1st came out. I thought it vindicated some of my own misgivings as to how people, including myself, thought. I was awed also by Jerome Bruner’s viewpoints.

          The problem today, from my own non-expert observation, is that the smartest people exhibit some of these biases routinely.

        • The 1st book I read by Bruner, when I was in 9th grade, was the Study of Thinking. I read it three or four times. That made me more attuned to how people categorized their thoughts, beliefs, expectations, prejudices, etc. I didn’t share my curiosity about thinking processes with others b/c I was so fascinated by how easily grownups would make hasty generalizations. When I pointed it out, I got an earful. So I just was content to listen to others intently.

          But the views that most intrigued me were in Bruner’s and Amsterdam’s Minding the Law. I thought that they did a brilliant legal and literary analysis of the Supreme Court racial discrimination cases and demonstrated implicitly & explicitly that in times of economic downturn, racial and ethnic animus are heightened. Judges & lawyers reflect their own beliefs biases; and shape norms by reifying notions of heroes and villians in culture. And in racial and ethnic characterizations, notions of heroes and villains are subtly interspersed to portray poorer minorities, in particular, as responsible for their economic lot in life.

        • Cool.

          My speculation is that academics in the past read and re-read what was on their bookshelves or on their nearby libraries’ often and very carefully. Maybe that which the disagreed with even more carefully.

          Recall CS Peirce being very critical of some of Laplace’s ideas e.g. Peirce’s retort that all clocks are clouds but have yet to read much of Laplace’s work outside the combination of observations stuff.

  3. Philip Tetlock & his wife Barbara Mellor through their Good Judgment Project [funded initially by US gov Intelligence Advanced Research Projects Activity {IARPA} have run tournaments for the intelligence community and the general public which test the potential for improved thinking and forecasting. So that strikes me as the one interesting project that should be followed. I particularly liked Philip Tetlock’s Political Expert Judgment. I was so relieved in fact that such a book came out.

    • Sameera: Yes, excellent book! Amazing! Reading this blog a person might forget that people can make good decisions without Bayes or NHST or any math at all.

  4. “Looking at decision making from a different direction, as Laplace’s faith in a clockwork universe that could be reduced to intelligible causes via the scientific method has been called into question with the discovery of quantum phenomena and emergent complexity, Laplace’s assumption that probability theory could serve as a domain-independent prescriptive model for human judgement has been upended by research demonstrating the relative efficacy of simple domain-specific decision rules and predictive models that respect cognitive limitations, tacit knowledge, multidimensionality of goals, and the need to adapt to complex and changing environments (Meehl, 1954; Gigerenzer and Brighton, 2009; Todd and Gigerenzer, 2000).

    It is not clear that the conclusion follows that Laplace has been upended, rather than the economists’ ideal of a rational individual. Unless you insist on a purely individualistic interpretation, it is not clear that groups (including the most experienced members of a the scientific community) suffer from the individual cognitive limitations. And it’s not even clear what a motivational bias means for a group. The scientific enterprise is not a heuristic, so I”m not quite sure how research, even experimental research, into heuristics and biases factor in.

    • Relatedly, despite diligent efforts, psychologists and management researchers have largely been unable to demonstrate that established organizations exhibit “implicit biases” in their decision making. “Explicit biases,” yes. But bureaucracies seem to be good at self-screening for unintended decision criteria.

      • Kyle

        I am not familiar with this research, and I am not sure what you mean by implicit bias, surely you don’t mean the technical term.

        In any event, if you mean the kinds of biases Laplace and the heuristics & biases program were talking about, I have seen plenty of examples of firms making decisions that fail to take into account of regression to the mean, that confuse performance with competence (the peter principle), that muck-up A/B tests due to over-confidence in small samples and other statistical errors. I am not saying its happening all the time—market selection is a thing—but its there.

        • Joshua, yes, I mean the technical term. You may know more about the literature than I do, but my reading is that efforts to identify implicit, unvoiced, correctable stereotypes affecting decisions in **organizational settings** has not panned out. People have plenty of explicit, voiced biases, of course, and discrimination surely exists, but this is not what the “implicit bias” researchers were looking for. I realize that this is different from heuristics. I thought my comment related to Steven Johnson’s comment that evidence does not necessarily show that professional communities exhibit the same cognitive limitations that individuals do.

        • If the question is whether “professional communities” or “bureaucracies” exhibit the same cognitive limitations people do, surely there is some overlap, surely some biases will go away or are attenuated, and surely some new biases will creep in. Not sure on the over-all pattern. In principle, without a lot of evidence, I would hesitate to label as biased the decisions of a practitioner with domain-specific expertise. That said, Laplace identified some errors in the collective common sense of the legal profession at the time.

          More to your point on implicit bias, my impression is that the experimental demonstrations of implicit bias haven’t panned out, but then their measures are pretty weak. To me it’s obviously true thing. For example, take something like accent; I don’t believe that Brits are on average smarter than Americans, but when I speak to a Brit my perception of their intelligence is automatic (i.e. unconscious, involuntary, implicit), and I have to actively correct for it. We know it goes the other way too. We know

          In the field, identifying whether discrimination is due to implicit bias vs. explicit bias is tricky. Most measures I have seen only get at whether or not unjustified discrimination exists (audit studies, subtle resume manipulations, and other innovative ideas etc.).

          I am sure it still happens in the field. I am sure when it comes to the interview stage for hiring/housing, etc. peoples biases come into play, and signals are interpreted differently based on priors, and people don’t realize it. Even straight-up statistical discrimination—i.e. not necessarily fair discrimination, but justified based on limited info—doesn’t have to be a conscious thing. Do you think professional baseball umpires realize that they are more likely to give better pitchers the benefit of the doubt on close pitches?

        • Concerning Brits and intelligence:

          (i) Presumably the sample of Brits you talk to is even more selected on intelligence than the sample of Americans you talk to.

          (ii) Even within the four British nations, there are many variants of English; common observation suggests that accent covaries with accent (more intelligent people tend to speak more posh). If someone talking to you spoke like the average Glasgow bricklayer, I wonder whether you’d automatically consider him more intelligent than would be appropriate (assuming you’d understand what he’s saying).

        • This is not a direct answer to your cogent commentary. As I understand it, Gordon Allport and many of his colleagues, while at Harvard and by serving on UN commissions inducted those minorities, a benign patronage of sorts, with the dual aim of attenuating racial & ethnic prejudice & engaging in leadership development here in US and abroad. Very ambitious endeavor. That was in keeping with the establishment of the UN and its broad mandate.

    • Hi Steven-

      We don’t think everything Laplace said has been upended, but there were two specific things he seemed believed in that were:

      1. clockwork universe, i.e. what we observe (noisily) is reducible to causes
      2. probability theory as general-purpose *prescriptive* model for decision making

      The two kind of go together.

      With regard to the cognitive limitations of groups, we didn’t consider that, but that is an issue too. Aside from group-think, the common sense of a profession can be in error or be leaving money of the table (think Moneyball), and Laplace gives several examples of carefully-reasoned judgments made by elite practitioners, and assented to by others, that were biased.

      I didn’t quite follow the connection between our paper and your mention of motivational bias, and the scientific enterprise as a heuristic.

      • Very interesting. I’m skeptical of the p(C|E) program yet hope to be converted by Pearl’s upcoming “The Book of Why”. My doubt comes honestly enough: through many years of watching courts struggle with causation. It started in law school when studying the famous case of Palsgraf v. L.I.R.R. There the dueling sides fought (thought they didn’t put it in the way I do here) over how to reconcile a clockwork universe with a vaguely understood notion of risk. Unable to resolve the problem they crafted a fig leaf called “foreseeability” and from there on all manner of claims were fought over the question of whether a particular outcome was foreseeable. By the way, it was from my realization that the “foreseeability” criterion hopelessly confuses those who judge in hindsight that I was led to Kahneman, Tversky, Gilovich, etc. looking for an answer to the problem of hindsight bias; and from there, after many dark paths and dead ends, to the following question.

        What some proposed alternatives to the causal framework when it comes to decision-making? When the courts began to deal with epidemiological evidence they seized upon the language of statistics and used it to turn uncertainty about the future into certainty about the past. In the quarter century that followed they ran with their new toy and used it to discover the causes of stock price drops, whether new medical procedures would have worked on patients who never had it, etc. (btw, reading judges’ interpretations of NNT – number needed to treat is a real horror show). Surely Statistics’ gravest sin was marketing a bunch of allegedly useful tools by taking ordinary words like error, power, confidence and significance, giving them bizarre new meanings, and then not telling anyone about it. Accordingly I similarly doubt that statistics has the answer. Andrew has referenced a proposal by Jimmie Savage a time or two but I’ve not been able to find it.

        • > taking ordinary words like error, power, confidence and significance, giving them bizarre new meanings, and then not telling anyone about it.

          As Sander pointed out earlier (if you read some of the link he gave on Bacon) that is a general hazard in all intellectual discourse, clearly identified by Bacon as the greatest nuisance – “Therefore shoddy and inept application of words lays siege to the intellect in wondrous ways” (Aphorism 43). Bacon considered these “the greatest nuisances of them all” (Aphorism 59). Because humans reason through the use of words they are particularly dangerous, because the received definitions of words, which are often falsely derived, can cause confusion.”

          An an aside, given I have answer someone’s query about an Exact test for linear trend today, I was thinking about the time I calculated Exact confidence intervals for relative risks for Peter Tugwell’s contribution to the US breast implant law suit around 2000 that you likely are familiar with. Now Exact methods try to make the Type one error rate of coverage rate closer to uniform than approximate methods. But in Epidemiological studies, because of confounding the approximate methods are almost surely very far from uniform and so efforts to make it a little less far have no at most negligible value. However, Peter’s group thought it was really important for me to calculate them to many decimal places.

        • Beautifully put: “Surely Statistics’ gravest sin was marketing a bunch of allegedly useful tools by taking ordinary words like error, power, confidence and significance, giving them bizarre new meanings, and then not telling anyone about it.”
          – Add to the list Fisher’s horribly mis-rationalized use of “null hypothesis” to refer to any hypothesis targeted for testing, instead of the ordinary English meaning of “no effect” – thus locking users into focusing on “no-effect” hypotheses as if no other could be tested.

          As Keith noted, Bacon saw the general problem 400 years ago. Yet statistics seems to remain adamant in its disregard if not contempt for ordinary English meanings and their importance for clear teaching and thinking.

        • Many fields (including the law!) use ordinary words to describe technical concepts. Those of us who operate in those fields need constant vigilance to see to it that we define our terms, point out the differences between “ordinary” definitions, and remind of this as needed. This is why, when I taught a continuing education course on “Common Mistakes in Statistics” I included in the first day (http://www.ma.utexas.edu/users/mks/CommonMistakes2016/WorkshopSlidesDay1_2016.pdf) topics such as “Terminology-inspired confusions” (p. 12), “Confusions involving the word “random” — Dictionary vs technical meanings” [pp.17ff, 31ff)

        • This must be Sander Greenland day on Twitter. I had no idea I was tweeting to one of your colleague Ken Rothman until he replied to me with reference to an article you co-authored. Then there is Charles SANDERS Pierce.

        • “Yet statistics seems to remain adamant in its disregard if not contempt for ordinary English meanings”

          I beg to differ. Humans in general – not just statisticians – rearrange the meanings of words to varying degrees to suit whatever needs they have at the moment to represent their own interests. Statistics does contribute, though: it provides numbers to associate with these misrepresented concepts, which is effective at convincing other people because a large share of humanity lacks the skill to deconstruct the language or the numbers and uncover the misrepresentation.

      • I did not understand that “general-purpose ‘prescriptive’ model for decision making” is to be construed as a failed heuristic, too complicated for practical use and too prone to error in its own right. But then, I was never clear that Laplace didn’t regard the clockwork universe as implying that a providential God was neither likely nor needed for a coherent universe to exist.

        If I understand it correctly one category of biases in the study of individuals is motivational, things like a halo effect. I don’t quite understand what the motivation of a group is. There are the explicit purposes for which groups are sometimes created, but in what sense is that unconscious?

        Laplacean probability analysis, like science, is not an algorithm, simple time saving procedure, intuitive approach, gut feeling, quick and dirty estimate, juryrigged solution. Which is why the study of heuristics doesn’t seem to be directly applicable to me, because those are what heuristics are…I thought.

        • Steven-

          thanks for the comment.

          When we wrote “general-purpose ‘prescriptive’ model” we were referring the Laplace’s probability theory, not any heuristic. Laplace viewed his probability theory as prescriptive, but his approach often involves reducing things to their constituent causes, which is not possible for many complex social phenomena. His clockwork universe-view did not account for this (not to mention quantum phenomena). There are other issues that prevent the theory from being prescriptive in many contexts, which we note.

          With regard to religion, while Laplace certainly took a dim view on the testimony relating to miracles, I don’t believe he took any position on the existence of God, behind that it was an unnecessary hypothesis.

          I am sure I am missing something, but I am not following where you are going with regards to groups. I don’t recall discussing the motivations of groups, or whether they are unconscious.

          thanks.

  5. Thanks for the paper, Joshua and Andrew. It is very refreshing to read historical takes on heuristics and biases, not from economics and psychology, but from mathematics.

    One of the responses uttered C.S. Peirce’ name in a different context but was wondering if Peirce had anything to say about chance and probability. Almost always, on all the important issues, he seems to have had a prescient intellect, anticipating important concepts in cognitive science, hence my curiosity.

    Also, I teach a course of heuristics and biases every year and every time I teach, I keep wondering if a more axiomatic approach that is more mathematical and something less statistical, and even less probabilistic is possible.

    Let me elaborate. Based on what we know about human judgment and decision making, agents don’t seem to be able to follow the *right* rules of probability and statistics. If that is the case, isn’t it reasonable to want to move away from probability and statistics and try to build a more predictive and explanative theory that is not based on probability but something else? Why should we expect to be able to explain human heuristics and biases in the language of probability and statistics? And hypothetically, if we had known about all these biases before axiomatic probability was invented, would we have had the same approach to heuristics and biases as we do today?

    PS: I know Geigerenzer has a few papers on the topic. And more notoriously, there is this misguided cottage industry of *quantum* cognition but I do not find them very satisfactory.

    • I have wondered whether it is feasible to move away from probability and statistics too. I speculate that with these Open Science initiatives [preprints, preregistrations, larger team collaboration, etc} that address qualitative/quantitative conundrums upon us, we will have a better idea of what next steps are.

      As for Charles Sanders Peirce, well it would require a very assiduous appraisal of his work, which I understand is scattered in various institutions and some of it still incomplete.

      • > very assiduous appraisal of his work, which I understand is scattered in various institutions and some of it still incomplete.

        That’s an understatement and likely the only one who ever got close was Frank P. Ramsey just before he passed away.

        On the other hand, Peirce: A Guide for the Perplexed by Cornelis de Waal may or may not give a digestible first impression of his overall work.

        • Keith I read Peirce’s Chance, Love, Logic a few years ago. I’ll take a look at Cornelis de Waal’s too. To digest Pierce would be a long haul I think Maybe some enterprising PHD candidate would undertake an assessment. I was curious about Pierce when I discovered that Dewey was one of his students at Johns Hopkins. I also have read a fair amount of William James work back in high school. I’m a product of many intellectual circles I think therefore. Pierce, in particular, is not easy reading. That is a hazard of much writing in the sciences & social sciences.

          As for the extent of Frank Ramsey’s acquaintance with Peirce. Relations among the thinkers of his time weren’t all that interesting despite the fact that they produced such influential work.

        • > Pierce, in particular, is not easy reading. That is a hazard
          This sets me up to relate Dewey’s response to a student who complained he was very hard to understand – “Young man you could make a very good career out of trying to understand me”.

          So not just hazard but opportunity if one perseveres (Peirce-everes). This presumes a confounding between difficult, new or subtle ideas with what is taken as poor writing. That is, I do think some too often take hard to understand to be mostly due to poor writing.

          (By the way, Dewey found Peirce too difficult as graduate student and did his best to avoid taking courses given by him but much later in his career did acknowledge Peirce as the main source of ideas in his book – Logic: A theory of inquiry.)

          > some enterprising PHD candidate would undertake an assessment
          Perhaps, though for me it was more a serious of short swims in frigid water with breaks in between (sometimes years) and (forced?) opportunities to apply the insights in my work. (The first was Chance, Love, Logic which I initially gave up on.) As time goes by, one can endure longer swims – but one likely would not choose to do so without finding the insights useful in their real work. I also had help from others in particular https://en.wikipedia.org/wiki/David_Savan and https://en.wikipedia.org/wiki/Ian_Hacking .

          Though I understand, we have to choose our windmills.

          > Relations among the thinkers of his time weren’t all that interesting
          Interesting is the eye of the beholder – I found his comments on Wittgenstein and Russel very interesting.

        • Peirce-everes! Witty. Just to clarify, I did not mean to single out Charles Peirce.

          Among the analytical writers of that period, Bertrand Russell was my favorite. Elegant writer. In fact, Russell could state Peirce’s arguments better than Peirce. That is a talent.

          I was curious from which article/book you read Peirce’s comments about Wittgenstein and Russell. I donated my Chance, Love, & Logic and some books on Frank Ramsey philosophy. So can’t cross check them. Were they from de Waal’s?

          Frank Ramsey’s sister, Margaret Paul has catalogued some of Frank Ramsey’s impressions. I gathered you may have meant Ramsey’s comments on Wittgenstein and Russell b/c Ramsey spent considerable time with both.

          Russell was more apt to go into great detail. Just anecdotes related to me by my father.

        • Facts and Propositions given to me when I was 15. Lastly F.P. Ramsey Critical Assessments by Maria J. Frapolli. I was checking for references to Charles Sanders Peirce specifically. Quite brief ones so far. Ramsey died at the age of 26. So he did not have as long to form relations. Bertrand Russell however did nice job drawing analytical distinctions between William James and John Dewy.

        • Well I discovered that Lettice Ramsey, F.P. Ramsey’s had a photographer’s shop in Cambridge, a name my my mother mentioned a few times. We lived in Cambridge England then. Good thing I was reviewing references to Charles S. Peirce

        • Ramsay’s unfinished book (focusing on Peirce’s work) has only recently being brought into focus – Cambridge Pragmatism From Peirce and James to Ramsey and Wittgenstein. Cheryl Misak https://global.oup.com/academic/product/cambridge-pragmatism-9780198712077?cc=ca&lang=en&#

          (Cheryl kindly shared a proof with me so the materials seems older to me.)

          > Russell could state Peirce’s arguments better than Peirce.
          I do think that is a misconception that Russel tried to correct very late in his career.

          Off into holiday mode – thanks for the comments.

        • Keith, Thank you. I haven’t come across that work. Mostly I have garnered some anecdotes from the academics that had known & written about the Cambridge Apostles or from my own immediate family/relatives. Some of them were their students at Cambridge.

          As I understand it, Russell took time to help writers at Cambridge. Was a unique atmosphere then too. Conducive to more eclectic path to an academic career. He lamented that so much was misinterpreted due to poor writing.

          I was schooled by some of Lettice Ramsey’s schoolteacher friends apparently. I recalled that last night. Girton Village, Cambridgeshire was small.

    • > I keep wondering if a more axiomatic approach that is more mathematical and something less statistical, and even less probabilistic is possible.

      Fuzzy logic?

      • Yes, that is often offered along with Dempster-Shafer as alternative modes of uncertainty quantification but they are do not offer a systematic induction machine to arrive at truth, unlike axiomatic probability and statistics combo. I would be interested to know how if the literature on Fuzzy logic and other formalisms have discussed heuristic and biases approach to understanding judgment and decision making.

        My hunch has always been hovering around the interface of computational complexity and learning theory but I’ve not made any progress probably because I have not learned theoretical computer science seriously and formally; I possess only a dilettante-ish knowledge of computer science, much like everything else in my life.

    • Rajesh:

      Not sure how to answer your query (see my response to Sameera) but I do recall Ian hacking once claiming that CS Peirce was the first highly accomplished philosopher to make chance and probability the center of their work.

      Now Peirce distinguished normative from descriptive and thought what he called the sciences of discovery needed to be normative – see here http://statmodeling.stat.columbia.edu/2017/11/29/expediting-organised-experience-statistics/

      On the other hand, and not shown in the diagram in the link, there are “Practical affairs” where Peirce repeated suggested custom and intuition should be one’s primary guide. Not sure I understand why but their was no doubt that all Peirce’s practical affairs ended in utter disaster.

      • Yes, I remember that post. In fact, I found the linked paper very interesting and filed it under *Peirce knew a lot* tag.

        I first encountered Peirce and American Pragmatism back in the day when I was in previous life as a young student of Physics reading about the history and foundations of statistical mechanics and thermodynamics. But since then, I’ve Peirce makes an appearance in my academic life frequently in machine learning, one could argue that Herb Simon’s work is part of the same lineage; mathematical philosophy, a philosopher from Newzealand who specializes in Peirce and the connection between discrete and continuum categories; philosophy of mind, via Dewey’s work on *externalist* conception of the mind.

        Peirce’s work is mysterious and I hope to read more of him and his work when I have some slack time.

    • If samples are large enough, probability and statistics can give you optimal decision rules. However, if they are small enough, the costs of variance overwhelm any gains from reducing bias.
      That’s why the statistical techniques work in experimental sciences, but have limited success in empirical/observational ones.

  6. People do not deviate from tenents of rationality when they are faced with uncertainty at all. They use optimal heuristics. In small samples, it’s grossly inefficient to try and reduce bias beyond certain level. Biased thinking is optimal.

Leave a Reply to Kyle C Cancel reply

Your email address will not be published. Required fields are marked *